jadesure / reinforcement-learning-based-rs

Reinforcement Learning Based RS by biclustering

This is the code reproduction for a paper: Reinforcement Learning based RS using Biclustering Technique. paper link It use a few techniques including data processing(pandas, numpy), biclustering algorithm (in the matlab), Q-learning (SARSA needs a little code change) and also data visulization with turtle and grid.
Results Demo click picture to see video in Youtube (ps: please ignore this picture, DQN which is not related to this project)

Unfortunately, the code reproduction is not success because too many details are not mentioned in the paper. It also includes a few theory flaws. For instance, there is no reason Unfortunately, the code reproduction is not a success because too many details are not mentioned in the paper. It also includes a few theory flaws. For instance, there is no reason why they map clusters into the grid by what manner, and how they do this. Thus, Q-learning is not convergency. The first author and relative supervisor are unavailable. (no reply with any email, no source code, no response)
If you are interested in this paper and have a good point of view, welcome to contact me :).
e-mail: [email protected]